Thanks to Richard Lipsey. I read the second edition of his book in undergraduate days and I am grateful for awakening a suspicion that there was something behind the eternal confusion with the reversal of axis. I have wrestled with this confusion for thirty plus years. 1. Marshall did not merely reverse the axis. It is a simple matter of convention whether one puts the independent variable on one or the other axis, and as such should not confuse anyone. Behind this axis reversal, Marshall seemingly killed economic science by aborting causation in two fundamental ways: by sabotaging epistemology and by perverting methodology. 2. In epistemology, the goal of science is to discover and articulate the causation, and Marshall did not pursue the study of causation. How could anybody ever even define demand without taking price as predetermined, far less let demand determine price? Can there be demand if the buyer has no idea what the price is? To suppose that demand determines price is a sabotage of epistemology. Price can never be set by one party: it must without exception be set by agreement between the buyer and the seller. One cannot represent the buyer with a demand curve and a seller with a supply curve at all: one must show the buyer with income and preference, and the seller with technology and endowments. Equality of demand and supply determines only the quantity, and cannot possibly determine the price. The price of x is the quantity of y that pays for x (per unit), and the equality of demand for and supply of x surely can in no way determine the quantity of y that pays for x. In short, the change in price is causally based on income from the buyers side and cost from the sellers side, and Marshall avoids both. Marshall presents a self-contradiction: price determines demand and supply, and then demand and supply determine price. His pupil Keynes was brighter, and got the idea that (aggregate) demand and (aggregate) supply determine output not price, and yet nobody noticed the reversal of Marshall by his pupil. 3. In methodology, Marshall violated the very old idea enshrined by the Buddha in the parable of six blind men trying to study an elephant. Partial equilibrium is bound to distort causality because it permits one to contradict the omitted but causally linked parts. The buyer must be a seller of something to earn the income with which he proposes to buy something, and that must not be left out as an exogenous variable, because that is very much an endogenous one. One can learn about neither price nor quantity unless one brings the production and consumption by each individual together in one full model: a partial model will never do. Thus Marshall is surely mistaken to suppose that an increase in demand can lead to an increase in price. It cannot. An increase in income can allow an increase in demand at a given price, or an increase in price for a given quantity, or a combination of both. If income is fixed, an increase is demand is possible only with a decrease in price. In consumer equilibrium, the price must be equal to lambda times the marginal utility of the good and the lambda is based on income. Without a change in income, no change in price is possible unless Marshall violates the equilibrium condition just mentioned: price must be equal to lambda times marginal utility. Just as the blind man who said the elephant was a curved and bony stick (after partially touching the tusk and ignoring the rest of the elephant), the partial model is necessarily distorted. Walras did it right with a general equilibrium model. The key culprit behind the Marshallian confusion is the very method of isolating the issue of determining variables, for leaving the casual variables out. An increase in demand cannot increase price: an increase in income can. Marshalls diagram of demand and supply of one good is a methodological absurdity: it is impossible for a single good to be bought and sold, because there must be at least two goods to pay for each other. Thus the demand for x must be made with an offer of supply of y and vice-versa in barter. Marshall got it wrong, and Walras got it right. With one good, there can be at most a numeraire to convert the quantity into value. The numeraire factor is not price. A single goods demand-supply intersection can at most determine its numeraire unit, not its price. The price must be a ratio. The so-called nominal price is a mistaken term to denote the numeraire value: it should not be called price. 2. Walras did it right to set up a general equilibrium model in which quantities are the dependent variables under predetermined or independent prices. Walras does not have a theory of price. He shows how the quantities are determined by equating the production and consumption for each good. His aim is to show that a unique set of prices would ensure that all quantities reached equilibrium, but he did not show how those prices would be reached. He left the work for us, and we failed mightily. The auction idea is meaningless in causal sense: the job is to describe the auction process. 3. Walras provides a valid starting point for a theory of price, but Marshall does not. To make any progress towards developing a theory of price, the first need is to throw away Marshall's mistake and self-contradiction. Here are the steps. First, take an autarkic individual who produces n different goods. There is no market and hence no market price. But optimization means that the subsistence producer must equalize the production of each good with its consumption, and that the marginal rate of substitution in production must be equal to the marginal rate of substitution in consumption for any pair of goods. The common rate of substitution in both production and consumption provides the equilibrium rate of substitution under autarky, and may be regarded as shadow price ratios. These are not market price ratios. 3. To get at market price (as a ratio), one must introduce entrepreneurship for profit. An optimizer does not seek profit, but merely allocates what he already has. An entrepreneur seeks profit and creates new value that did not exist before. He can do this by exchange such that the buyer offers a price higher than the marginal cost of the producer but lower than the marginal opportunity cost of the same good if the buyer were to produce the good. To formalize this, one must add another set of equations to the Walrasian model. Those equations must show that in each transaction, the value of the first good which is paid for by the second good is equal to the value of the second good. The price of the first good in the market then is the quantity of the second good per unit of the first good. For the seller of the first good, the quantity of the second good received in payment must be higher than the quantity of the second good he could produce in substitution of the first; for the buyer of the first good, the quantity of the second good he pays out must be lower than the quantity of the second good he could produce in substitution if he produced the second good instead of the first. The price then must be settled by bargaining between the buyer and the seller. It must be higher than the sellers production cost and lower than the buyers production cost, both measured in terms of the other good. The same argument can be repeated with marginal utility in place of marginal cost. 4. To put it in gist, the optimizer is a price taker who chooses only the quantities and not the price, while the entrepreneur is someone who does not choose quantities, but prices. It is as if producers presume some expected price, say from their sense of immediate past, and choose quantities optimally. Then they go to the market and become entrepreneurs. They already have the quantities, and now they must choose prices to set their values such that the goods that pay for each other become equal in value, namely, such that a certain quantity of the second good pays fully and fairly for a unit of the first good. Walras can be and must be retained, but Marshall, NAUGHT. Price theory must presume that the quantities have already been chosen and have arrived in the market. A dynamic model will of course show what happens in the next round of quantity choices after the change in price by todays market clearing. 5. The foundation of price theory is the idea of equivalence, namely that the two goods that pay for each other must have equal value. Price is the ratio to establish this equivalence. One must go beyond Walras to show the equivalence. One can see the essential validity of classical price theory if one thinks of value as the weight of demand and supply. Thus suppose that at the price of previous day, the value of todays demand is larger than the value of supply. We can visualize a scale in which one side takes the demand and the other side takes the supply and the scale must come to a balance. Now, if the previous days price makes demand heavier than supply, the rise in price will increase the weight (value) of supply (and with some reduction in quantity from the demand side) will reduce the weight of demand until balance is established. In the abstract, this mechanical process seems to tell the story right. However, we must tell the human story of the buyers and sellers as entrepreneurs. This means that instead of a machine automatically balancing demands with supplies in terms of value by adjusting prices, we must show the human process of bargaining. Of course one kind of bargain may appear like auction, but an auction as such does not fully reflect the price-setting process. We must have the arbitrageur here. How come we have too much of Marshall and too little of Walras? Mohammad Gani